Resources / Governance Toolkit / Checklists

CHECK EXISTING TOOLS

Meeting Transcription Checklist

Meeting transcription is where shadow AI creates the most risk. Meetings contain sensitive discussions -- strategy, personnel decisions, client data, financials -- and transcription tools record all of it. Before adding any tool, audit what you already have. You probably have more capability than you think.

Evaluation Checklist

1. Audit What You Already Have

Check your meeting platform's built-in AI features

Microsoft Teams

Copilot in Teams provides real-time transcription, meeting summaries, action items, and post-meeting recap. Available with M365 Copilot license or Teams Premium. Transcripts stay within your Microsoft 365 tenant. If you're already paying for M365 Copilot, this is included.

Google Meet

Gemini in Meet provides transcription, meeting notes, and summaries. Available with Workspace Business Standard and above. Notes are saved to Google Docs automatically. If you're on Google Workspace, check if your plan includes this -- it may already be active.

Zoom

Zoom AI Companion provides meeting summaries, smart chapters, and next steps. Available on paid Zoom plans at no additional cost. Transcription is built in. Companion can draft follow-up emails and identify action items.

Webex

Webex AI Assistant provides real-time transcription, meeting summaries, and catch-up features. Available on Webex Suite plans. Check your current licensing.

Run a comparison test

Before evaluating third-party tools, test your existing platform's transcription on 3-5 real meetings:

  • - Is the transcript accurate enough for your needs?
  • - Does the summary capture the right key points?
  • - Are action items identified correctly?
  • - Can you search past transcripts effectively?
  • - Does it work for your typical meeting types (1:1, standup, all-hands, client calls)?

If it scores well on most criteria: You don't need a third-party tool. Enable the feature for your organization, provide brief training, and move on.

2. Can Your General AI Assistant Fill the Gap?

If your meeting platform falls short, try your General AI Assistant next

Most General AI Assistants can process audio or transcript files effectively:

  • - Record the meeting using your platform's built-in recording
  • - Export the transcript (or audio file)
  • - Upload to your General AI Assistant for summarization, action item extraction, or analysis

This is a manual workflow, but it avoids adding another tool. It works well for teams that only need post-meeting summaries occasionally rather than real-time transcription for every meeting.

Best for: Teams with moderate meeting volume who don't need live transcription.

Not ideal for: Teams that need real-time captions, automated recording of all meetings, or searchable archives across hundreds of meetings.

3. Evaluating Third-Party Transcription Tools

Only get here if your existing platform and General AI Assistant both fall short.

Options comparison

Otter.ai Business

Real-time transcription with speaker identification. Integrates with Zoom, Teams, and Google Meet. OtterPilot joins meetings automatically. Searchable transcript archive. SOC 2 Type II certified. Per-seat subscription.

Best for: Teams that need automated transcription across multiple meeting platforms with a searchable knowledge base.

Fireflies.ai

AI meeting assistant that joins calls automatically. Transcription, summaries, action items, and analytics. Integrates with CRM systems (Salesforce, HubSpot). Topic tracking across meetings. SOC 2 Type II certified. Per-seat subscription.

Best for: Sales teams and organizations that need CRM integration and meeting analytics/intelligence.

Key evaluation criteria for third-party tools

Data Processing Agreement (DPA)

Where is transcript data stored? How long is it retained? Can you control retention periods? Is data encrypted at rest and in transit? Can you delete transcripts on demand? These tools ingest everything said in a meeting -- treat them with the same security scrutiny as your most sensitive data systems.

Training on Your Data

Verify that the tool does not train models on your meeting transcripts. Both Otter.ai Business and Fireflies.ai commit to not training on business plan data, but confirm this in writing for your specific agreement.

Bot Presence Notification

Most transcription tools join meetings as a visible "bot" participant. This serves as implicit notification that the meeting is being recorded. Check your organization's recording consent policies -- some jurisdictions require explicit consent from all participants.

Meeting Type Filtering

Not every meeting should be transcribed. Look for tools that let you set rules: auto-record external client calls but skip 1:1s, or transcribe team standups but not HR conversations. The inability to filter meeting types is a common source of friction.

4. Recording & Consent Policies

This is a policy decision, not just a technology decision.

Before deploying any transcription tool, work with Legal and HR to establish clear policies:

  • Which meeting types are recorded? All meetings, external calls only, opt-in per meeting, or specific categories? Most organizations find that "all meetings" creates pushback. Start with opt-in and expand based on adoption.
  • Who has access to transcripts? Only attendees? Managers? The entire organization? Set clear access controls before launch, not after.
  • How long are transcripts retained? 30 days, 90 days, 1 year, indefinitely? Align with your existing data retention policies. More data isn't always better -- it's also more liability.
  • How are external participants notified? Client calls and vendor meetings have different consent requirements. Ensure your tool provides clear notification to all participants.
  • What about sensitive meetings? HR discussions, legal matters, board meetings, and executive strategy sessions may need to be explicitly excluded. Build this into the policy from day one.

Decision Flow

Use your existing meeting platform if:

  • - You're on Teams (with Copilot), Google Meet (Workspace Business+), or Zoom (paid plan)
  • - Your test of built-in transcription meets accuracy and summary needs
  • - You don't need cross-platform transcription or CRM integration
  • - You want to avoid adding another tool with access to all meeting content

This covers the majority of organizations. Enable the feature, train your team, done.

Use your General AI Assistant if:

  • - You only need post-meeting summaries (not real-time transcription)
  • - Volume is low (a few meetings per week that need summarization)
  • - You want to avoid giving another tool access to meeting recordings

Add a dedicated tool (Otter.ai / Fireflies.ai) if:

  • - Your meeting platform's built-in features are insufficient after testing
  • - You need cross-platform transcription (meetings happen on multiple platforms)
  • - CRM integration is a requirement (sales teams, customer success)
  • - You need a searchable archive across hundreds of meetings
  • - Meeting analytics and intelligence are a business priority

Shadow meeting transcription is your highest-risk shadow AI category.

If employees are already using personal Otter.ai or Fireflies.ai accounts for work meetings, all of that meeting data is sitting in a consumer account with no enterprise data protections. This is the first shadow AI category to address during your governance rollout -- don't wait until Week 3. If your amnesty audit reveals widespread personal transcription tool usage, prioritize providing an approved alternative immediately.

Previous: Image Generation Back to Governance Toolkit